Technology providers are banking on generative AI to accelerate revenue gains. They are planning accordingly.
“The next industrial revolution has begun,” Nvidia President and CEO Jensen Huang said during a May earnings call. The company most closely associated with graphics processing unit technology saw revenues balloon during the first half of the year, driven largely by demand for its AI-optimized silicon.
“For every dollar spent on Nvidia AI infrastructure, cloud providers have an opportunity to earn five dollars in GPU instant hosting revenue over four years,” Huang said. “The CSPs are consuming every GPU that's out there.”
The gains are predicated on a flurry of enterprise tech spend that has only begun to materialize. But the building boom is already under way.
As vendors embed models in their product portfolios, hyperscalers and chipmakers are pouring capital into infrastructure to accommodate an impending demand surge, packing data centers with servers optimized for high-capacity workloads.
AI-related data center investments are expected to push global infrastructure spending to $1 trillion by 2028, according to Dell’Oro Group. The market research firm revised its infrastructure spend forecast upward Thursday as a result of “surging demand in AI-related data center infrastructure.”
The firm now expects 24% compound annual growth in the data center market over the next five years, the report said.
Cloud providers have to lay the groundwork for AI before they can reap the rewards Huang foresees.
“The reality right now is that while we're investing a significant amount in the AI space and in infrastructure, we would like to have more capacity than we already have today,” Andy Jassy, Amazon president and CEO, said Thursday during the company’s Q2 2024 earnings call.
Amazon's cloud division saw revenues grow 19% year over year to $26.3 billion. The company spent $30.5 billion in capital expenditures, mostly building out AWS infrastructure capacity during the first half of the year, and expects to invest more during the remaining six months, SVP and CEO Brian Olsavsky said.
Jassy acknowledged AI is still in its early days but said AWS has already secured “a multibillion-dollar run rate.” He also laid out the challenges inherent in massive infrastructure projects.
“If you end up actually with too little capacity, then you have service disruptions, which really nobody does because it means companies can't scale their applications,” Jassy said. “However, if you actually deliver too much capacity, the economics are pretty woeful.”
Cloud’s big 3
The three largest hyperscalers are fighting for shares of an expanding market. Their relative positions have remained relatively stable over the last several years, even as generative AI has spurred enterprise migrations.
Amazon held on to nearly one-third of the global cloud market during Q2, according to Synergy Research Group. Microsoft saw its share drop to 23% while Google gained a percentage point to land at 12%, the firm said in a Thursday report.
The total cloud infrastructure and platform services market grew 20% year over year for the third consecutive quarter, SRG said.
Hyperscaler market share shifted marginally in Q2 as AWS remained the largest global cloud provider
As Microsoft and Google grow hyperscaler empires, they are just as focused on infrastructure capacity expansion as AWS.
Microsoft had a $19 billion capital expenditure bill for the three-month period ending June 30, the company’s EVP and CFO Amy Hood said Tuesday, during the company’s Q4 2024 earnings call. AI-related investments in Azure infrastructure and third-party data center leases accounted for nearly all of the spending, she said.
The company saw Azure revenues increase 29% year-over-year revenue growth and attributed eight of those percentage points to AI services, where demand was higher than available Azure compute capacity, Hood said. Microsoft will continue to ramp up data center investments during the second half of the year, she added.
Google’s infrastructure investments eclipsed its cloud revenues during Q2. The company spent $13 billion on technical infrastructure, including servers and data centers, and reported $10 billion in quarterly revenues — a record for the junior hyperscaler.
Alphabet CEO Sundar Pichai defended the company’s Google Cloud infrastructure strategy during the company’s Q2 2024 earnings call last month.
“The risk of under-investing is dramatically greater than the risk of over-investing,” Pichai said. “Even in scenarios where it turns out that we are over-investing, these are infrastructure which are widely useful for us.”
Beyond cloud
The data center spending spree has spread beyond cloud to model builders and chipmakers, too. Facebook parent company Meta's capital expenditures reached $8.5 billion, mostly invested in servers, data centers and network infrastructure to support its Llama large language models during the second quarter.
The company announced plans to build $800 million data centers in Alabama, Wyoming and Indiana during the first half of the year.
Meta CEO Mark Zuckerberg echoed Pichai’s sentiments during the company’s Q2 2024 earnings call. “At this point, I'd rather risk building capacity before it is needed, rather than too late, given the long lead times for spinning up new infra projects,” he said.
Meta joined an AI infrastructure open-standard alliance of big tech firms, including AMD, Broadcom, Cisco, HP, Intel, Google and Microsoft, in May. The Ultra Accelerator Link group aims to develop technical specifications to ease data center interconnectivity.
“The industry is coming together to establish a standardized approach for building the next generation of high-performance data center AI solutions at scale,” AMD President and CEO Lisa Su said during the company’s Q2 2024 earnings call Tuesday.
The chipmaker saw its data center revenues increase 115% year over year to a record $2.8 billion in Q2, driven by hyperscaler demand for its silicon.
Su expects infrastructure spending sprees to continue. Her company is counting on it.
“We have to invest,” Su said. “The potential of AI is so large.”
